The Superagent and the Resolver take each problem as far as it needs to go.
The Superagent is Plannk's AI agent for business. Day to day, it works with the web, sector databases, your project's files and your connected tools. For a hard decision, the Resolver builds the complete study, one workstream at a time.
You ask the Superagent what you would ask someone on the team.
It could be comparing two contracts, pulling together the numbers for a market, building the budget spreadsheet or writing the memo.
- Contracts, spreadsheets, meetings and earlier research go into the answer without you sending anything again.
- Sector databases and sites that need real navigation fill in what the open web does not carry.
- The result comes back as a document, spreadsheet, chart, image or map, and stays editable in the conversation.
An opinion that takes two to three weeks reaches the board in two to three days with Plannk.
The Rondonópolis Rail Spur licensing, public at Brazil’s IBAMA, runs to 7,753 pages. The Resolver returned the opinion on its 24 license conditions with the page for each, and the team reviews and signs it.
Open the full opinion. Opens in a new tab.| The traditional wayselective reading of the record | With a chat assistantChatGPT, Claude or Copilot | In Plannkone request to the Resolver, 49 min of execution | |
|---|---|---|---|
| Time to decision | 2 to 3 weeks | 2 weeks | 2 to 3 days |
| Total cost, estimate | about R$15,0003 people, team R$15,000, AI and tools R$0 | about R$10,0003 people, team R$10,000, AI and tools R$260 | about R$1,8001 to 2 people, team R$1,700, AI and tools R$100 |
| Team hours | about 140 h | about 90 h | about 15 h |
Hours by stage
| Stage | The traditional way | With a chat assistant | In Plannk |
|---|---|---|---|
| reading the record | 86 h | 43 h | no hours |
| license conditions | 24 h | 15 h | no hours |
| drafting | 16 h | 9 h | no hours |
| request and coordination | no hours | 11 h | 1 h |
| review and sign-off | 14 h | 13 h | 14 h |
Conservative estimate: selective reading, R$111 an hour with payroll charges (median of the Robert Half 2026 Salary Guide. Opens in a new tab.), the chat assistant’s highest published gain, and AI priced by time of use or by the opinion’s credits. Measured: the execution and the size of the record.
For a hard problem, the Resolver splits the request into workstreams and closes each one with a source and a calculation.
The decision is whether to pursue the Gulfport concession. Each workstream answers part of the question, and the port records change the premise.
- Port records 2019–2025
- Cargo profile
- Data reconciliation
- Scenarios to 2040
- Post-Katrina rebuild
- Berth lease to 2043
- Authority accounts
- CAPEX benchmarks
- Expansion sizing
- Yilport precedent
- Federal permits
- Hurricane risk
- Lease and P3 deals
- Gated recommendation
USACE
Corps of Engineers and federal container counts · 2019–2025
- 2019–2025
- Demand
- Investment
- Port authority statute, accounts and lease precedents
2026–2040
- Conservative
- Base
- Optimistic
Three scenarios for the terminal
The only container berth was already leased through 2043, with options to 2063. The study sized the demand, priced three investment scenarios and recommended holding capital while watching the triggers that would reopen the decision.
The only container berth was already leased through 2043, with options to 2063. The study sized the demand, priced three investment scenarios and recommended holding capital while watching the triggers that would reopen the decision.
USACE and port authority data, 2019–2025. Scenarios through 2040.
- Reframing the question as a capital decision.
- Opening fourteen workstreams.
- Crossing Corps of Engineers and federal container counts.
- Projecting three scenarios to 2040.
- Reading the statute and accounts of the port authority.
- Reviewing text, spreadsheets and charts.
- The study is ready to inspect.
The leadership team asks the Superagent what changed in each project.
The answer crosses meetings, tasks, deadlines and studies from every project you follow. The context arrives ready, with no status meeting, and the conversation with the team starts from the decision.
Frequently asked questions
How does the Superagent compare with ChatGPT or Claude?
In a chat assistant, you run the work: you split the question, find the data, check the math and assemble the file across several conversations in a project. The Superagent queries sector databases beyond the open web and starts from what the project has already produced, from meetings to spreadsheets. A hard problem goes to the Resolver, which returns the study: in the port case, the editable document arrived in two and a half hours. Plannk keeps up with the evolution of AI models and uses, at each step, the one that gives the best result for the cost, without your team keeping a subscription for each model.
When do I use the Superagent and when do I take the request to the Resolver?
The Superagent handles what a conversation can settle: a question, a calculation, reading a contract, a spreadsheet or a short document. The Resolver takes the decision that needs a complete study, with several workstreams, sources and calculations, and works through to the recommendation. Both start from the files of the same project.
Does the Superagent see every project in the company?
It sees the projects you can see: your own, the ones whose team you belong to and the ones shared with the organization, always under each project's permissions. In them it reads meeting minutes and action items, tasks, deadlines, budget and studies, and answers about several projects in a single question. A routine can prepare that summary every Monday and email it to up to ten people.
Do I need a complete brief before I start?
No. Describe the decision the work has to support and what needs to be ready at the end, including when that means several files, such as a study, a spreadsheet and a presentation. The Resolver settles audience, constraints, sources and format in the conversation before moving on.
What happens before the study reaches me?
The Resolver checks its own work before returning it: it redoes the calculations behind the conclusion, checks every cited source and corrects what does not hold up, in text, spreadsheets and charts. The final technical review stays with you.
Can I close the page while the Resolver is working?
Yes. Once the study starts, it continues with the page closed. When you return, open the conversation to follow or review it; use Stop to cancel the work in progress.
What happens when my files disagree with an external source?
Say which document, period or source counts as the main reference. If the priority is unclear, the conflict stays visible in the answer or the study, and the decision stays with you.
When does a revision still belong to the same request?
Adjustments to content, structure, depth and presentation continue in the same conversation. If the decision, the audience or the scope changes completely, a new request keeps the reasoning and the files of each piece of work separate.
One request unblocks the decision that is waiting for a study.
Day to day, the Superagent answers with what your projects already hold. When the decision needs a complete study, say what needs to be ready at the end, and the Resolver returns the study for you to review.